| Message ID | 1387189568-31769-1-git-send-email-fan.du@windriver.com |
|---|---|
| State | Changes Requested, archived |
| Delegated to: | David Miller |
| Headers | show |
On Mon, Dec 16, 2013 at 06:26:08PM +0800, Fan Du wrote: > > And also since xfrm_policy_sk_bundles is only used in xfrm_lookup and > __xfrm_garbage_collect, both in process context, no reason we should turn > BH off. Are you sure about that? __xfrm_garbage_collect() is called via dst_alloc() which can be called from softirq and process context. > In addition we can use xchg to avoid the spinlock, inspired by > discussion in: http://marc.info/?l=linux-netdev&m=138713363113003&w=2 > > Signed-off-by: Fan Du <fan.du@windriver.com> > --- > Please note this patch is based on commit 283bc9f35bbbcb0e9ab4e6d2427da7f9f710d52d > ("xfrm: Namespacify xfrm state/policy locks"), which is still in ipsec-next tree. > You could use Subject: [PATCH ipsec-next] xfrm: Namespacify xfrm_policy_sk_bundles then everybody should know on which tree it is based. -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 2013年12月16日 20:23, Steffen Klassert wrote: > On Mon, Dec 16, 2013 at 06:26:08PM +0800, Fan Du wrote: >> >> And also since xfrm_policy_sk_bundles is only used in xfrm_lookup and >> __xfrm_garbage_collect, both in process context, no reason we should turn >> BH off. > > Are you sure about that? > > __xfrm_garbage_collect() is called via dst_alloc() which can be called > from softirq and process context. Thanks for pointing this out, you are correct! :) IMO, xchg can still cover those two cases: 1. xfrm_lookup(Process context) vs __xfrm_garbage_collect(softirq context) 2. xfrm_lookup(Process context) vs __xfrm_garbage_collect(Process context when SPD change or dev down) I will fix commit message properly on v2 if you are ok with above description. >> In addition we can use xchg to avoid the spinlock, inspired by >> discussion in: http://marc.info/?l=linux-netdev&m=138713363113003&w=2 >> >> Signed-off-by: Fan Du<fan.du@windriver.com> >> --- >> Please note this patch is based on commit 283bc9f35bbbcb0e9ab4e6d2427da7f9f710d52d >> ("xfrm: Namespacify xfrm state/policy locks"), which is still in ipsec-next tree. >> > > You could use > > Subject: [PATCH ipsec-next] xfrm: Namespacify xfrm_policy_sk_bundles > > then everybody should know on which tree it is based. > >
On Tue, Dec 17, 2013 at 09:45:46AM +0800, Fan Du wrote: > > > On 2013年12月16日 20:23, Steffen Klassert wrote: > >On Mon, Dec 16, 2013 at 06:26:08PM +0800, Fan Du wrote: > >> > >>And also since xfrm_policy_sk_bundles is only used in xfrm_lookup and > >>__xfrm_garbage_collect, both in process context, no reason we should turn > >>BH off. > > > >Are you sure about that? > > > >__xfrm_garbage_collect() is called via dst_alloc() which can be called > >from softirq and process context. > > Thanks for pointing this out, you are correct! :) > > IMO, xchg can still cover those two cases: > 1. xfrm_lookup(Process context) vs __xfrm_garbage_collect(softirq context) > 2. xfrm_lookup(Process context) vs __xfrm_garbage_collect(Process context when SPD change or dev down) > > I will fix commit message properly on v2 if you are ok with above description. Yes, please fix and resend. -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/include/net/netns/xfrm.h b/include/net/netns/xfrm.h index 1006a26..4a30b1b 100644 --- a/include/net/netns/xfrm.h +++ b/include/net/netns/xfrm.h @@ -58,9 +58,9 @@ struct netns_xfrm { struct dst_ops xfrm6_dst_ops; #endif spinlock_t xfrm_state_lock; - spinlock_t xfrm_policy_sk_bundle_lock; rwlock_t xfrm_policy_lock; struct mutex xfrm_cfg_mutex; + struct dst_entry *xfrm_policy_sk_bundles; }; #endif diff --git a/net/xfrm/xfrm_policy.c b/net/xfrm/xfrm_policy.c index a7487f3..26d79c0 100644 --- a/net/xfrm/xfrm_policy.c +++ b/net/xfrm/xfrm_policy.c @@ -39,8 +39,6 @@ #define XFRM_QUEUE_TMO_MAX ((unsigned)(60*HZ)) #define XFRM_MAX_QUEUE_LEN 100 -static struct dst_entry *xfrm_policy_sk_bundles; - static DEFINE_SPINLOCK(xfrm_policy_afinfo_lock); static struct xfrm_policy_afinfo __rcu *xfrm_policy_afinfo[NPROTO] __read_mostly; @@ -2108,12 +2106,8 @@ struct dst_entry *xfrm_lookup(struct net *net, struct dst_entry *dst_orig, } dst_hold(&xdst->u.dst); - - spin_lock_bh(&net->xfrm.xfrm_policy_sk_bundle_lock); - xdst->u.dst.next = xfrm_policy_sk_bundles; - xfrm_policy_sk_bundles = &xdst->u.dst; - spin_unlock_bh(&net->xfrm.xfrm_policy_sk_bundle_lock); - + xdst->u.dst.next = xchg(&net->xfrm.xfrm_policy_sk_bundles, + &xdst->u.dst); route = xdst->route; } } @@ -2551,11 +2545,7 @@ static void __xfrm_garbage_collect(struct net *net) { struct dst_entry *head, *next; - spin_lock_bh(&net->xfrm.xfrm_policy_sk_bundle_lock); - head = xfrm_policy_sk_bundles; - xfrm_policy_sk_bundles = NULL; - spin_unlock_bh(&net->xfrm.xfrm_policy_sk_bundle_lock); - + head = xchg(&net->xfrm.xfrm_policy_sk_bundles, NULL); while (head) { next = head->next; dst_free(head); @@ -2942,7 +2932,6 @@ static int __net_init xfrm_net_init(struct net *net) /* Initialize the per-net locks here */ spin_lock_init(&net->xfrm.xfrm_state_lock); rwlock_init(&net->xfrm.xfrm_policy_lock); - spin_lock_init(&net->xfrm.xfrm_policy_sk_bundle_lock); mutex_init(&net->xfrm.xfrm_cfg_mutex); return 0;
xfrm_policy_sk_bundles, protected by net->xfrm.xfrm_policy_sk_bundle_lock should be put into netns xfrm structure, otherwise xfrm_policy_sk_bundles can be corrupted from different net namespace. And also since xfrm_policy_sk_bundles is only used in xfrm_lookup and __xfrm_garbage_collect, both in process context, no reason we should turn BH off. In addition we can use xchg to avoid the spinlock, inspired by discussion in: http://marc.info/?l=linux-netdev&m=138713363113003&w=2 Signed-off-by: Fan Du <fan.du@windriver.com> --- Please note this patch is based on commit 283bc9f35bbbcb0e9ab4e6d2427da7f9f710d52d ("xfrm: Namespacify xfrm state/policy locks"), which is still in ipsec-next tree. --- include/net/netns/xfrm.h | 2 +- net/xfrm/xfrm_policy.c | 17 +++-------------- 2 files changed, 4 insertions(+), 15 deletions(-)